SPL Meter Correction Tables - Page 3

I did reply to both your comments. I'm just not familiar enough and don't really understand C-Weighting well enough to chisel in stone a comment about it. My statement included an "I also didn't think" preface and this comment at the end: "Maybe Ilkka could elaborate because I may not fully understand myself." My thinking ain't always accurate and I believe he would be more familiar with this than I would. I have seen comments by some of the more knowledgeable to this effect, or at least I thought they were more knowledgeable.

Without a doubt though, I do know that by using C-Weighting it does NOT make the necessary corrections for the RS mic to be accurate... you DO need correction values for all of the RS SPL meter mics.

I have an older model analog and 3 newer models. I have tested between the older and two of the newer models and they were all three different. It wasn't by much but it was different.

The whole key here is that the old correction values are not adequate... they are way off for any meter I have tested against my calibrated ECM8000. However using the newrs.cal file that John Mulcahy, Ken Bruce and myself created, it will get you closer than any other .cal files I have tested. Due to the variances from meter to meter, you may be off a couple of db but that is much better than being off as much as the old correction values would cause you to be off.

Many moons ago, before I was actually a member here... I say about 5 years ago, there was a guy here on AVS that posted his results of testing the RS Meter and stated the old correction values were accurate... therefore I always used them. I just simply decided to acquire a better mic and have it professionally calibrated so that I wouldn't have to use the RS mic. I decided to test the differences and learned how drastic a difference there is. I came here and noticed Ilkka had done the same.

I posted this above in an earlier post showing the differences between his test and my test:

10 Hz +5 dB vs. +27.00
12 Hz +6 dB vs. +21.91
16 Hz +14.8 dB vs. +16.15
20 Hz +11.5 dB vs. +12.38
25 Hz +7.8 dB vs. +8.80
30 Hz +5 dB vs. +6.50
40 Hz +3.2 dB vs. +3.46
50 Hz +2 dB vs. +2.21
80 Hz +1 dB vs. +0.98
100 Hz +1.2 dB vs. +0.27

We are within 1.5db of each other from 16hz to 100hz which would be a very reasonable difference from one RS meter mic to the next. What I would question is possibly his calibrated mic's accuracy at 10hz and 12hz. I have a problem believing that any RS meter mic is only -5db at 10hz or -6db at 12hz. Our correction values would be more rational when considering what the old ones were... according to the old corrections, 10hz was down 20db... we adjusted it to 27db. At 12.5hz it was down 16.5db and we adjusted it to ~21db.... while he went the opposite direction with his adjustments.

That's about all I know to tell ya.

I'm not sure what happened to Ilkka... I hope he's okay. I've sent him a few PM's in various forums but he's not responded and I noticed he hasn't posted here since March... which is highly unusual for him being he's made almost 2200 posts since Feb 2005.

Btw... I have two brand new analog RS Meters sitting on my table. I plan to measure these against my calibrated mic. I will not need them when finished and will be letting them go for exactly what I have in them and they will include the calibration file with them. I am simply doing this to test between meters. Hopefully I'll get around to this in the next few weeks.

Hi again, Sonnie,

Thanks for the reply again. I know that you had responded to my previous post but I still felt that the question about whether the RCA output of the RS digital SPL meter was c-weighted compensated or uncompensated was still unresolved, that is, not tested and verified by anyone else other than the forum member, Ethan Winer, of whose results that Ilkka posted which suggested that the RCA output was uncompensated unlike the RS analog meter output which is C-weighted compensated.

I believe this would make a big difference when using the REW program and the RCA output of the RS digital SPL meter because according to Ilkka on post #16 of this thread, it says that the value of +6.2 dB at 20 Hz (as taken from a previous compensation table that he had come up with which is referenced in post #10 of this same thread) comes from the general c-weighting compensation. Note this value he is talking about is not referring to any SPL meter inaccuracy or otherwise but is referring to the general c-weighting compensation definition. So I take it to mean that if one had a perfectly flat microphone at 20 Hz but one took the output of that mic at 20 Hz and applied the c-weighting compensation to the output (like the RCA output of the analog RS SPL meter does, that is, it applies a c-weighting compensation), then the output would decrease by 6.2 dB so one would have to add back 6.2 dB to the reading to get it back to normal.

Now the correction values you have come up with were done with the RCA output of the analog RS SPL meter which is known and verified to have a c-weighting compensation done to it so your values corrects for the both the c-weighting compensation plus the inaccuracy of the RS SPL meter mic. So using your value of +11.5 dB for 20 Hz, I would take this to mean that the microphone was inaccurate by -5.3 dB and then the c-weighting compensation of -6.2 dB was added to it by the c-weighting compensation circuits of the RS analog SPL meter so if you add these two numbers together you get your -11.5 dB deviation that you came up with.

Now with the digital RS SPL meter, if it is the case, as Ethan Winer had supposedly verified, does not have the c-weighting compensation applied to the RCA output, then even if the microphone from your analog RS SPL meter was moved into the digital RS SPL meter, since the RCA output does not supposedly have the c-weighting compensation, the correction value at 20 Hz would only be +5.3 dB using your findings.

This is why I think it is important that someone else test another one or two or more digital RS SPL meter RCA outputs and see if Ethan Winer is correct or not.

P.S. I found a webpage here: http://www.diracdelta.co.uk/science/...ng/source.html that gives the formula for c-weighting compensation plus has a calculator that you can input any frequency and these values follow exactly the compensation table of Ilkka's that is referenced in post #10 of this thread.

What you say makes good sense. I think if you will post a question about this in the BFD | REW forum (link in sig), you might can get John Mulcahy, Ken Bruce, Wayne Pflughaupt, all to chime in on this. I believe you'd get some answers maybe... John understands this stuff pretty well... a lot better than I do.

I can tell you that when I did my testing on the analog units I did not compensate for C-Weighting in REW.

I would suggest Ethan knows what he's doing and would be correct with his testing, however, whoever manufactures this meter could have made some changes, therefore further testing may be beneficial. Since I have two analog meters to test, I'll see if I can get my hands on a couple of digital meters as well and test them too.


EDIT: Have two digital meters on their way. I will test one older analog, two new analog and two new digital meters all at the same time and report my findings in the BFD | REW forum once concluded.

The results are in... whether C-Weighting is applied or not is irrelevant... the meters (all of them) still need correction values applied.

See the test here!

And the resulting calibration files here.

Sonnie has produced calibration files for the old analog meter (which we already had), the new analog meter, and the (only) digital meter - the first ones I've seen for the latter two. Note that the digital meter is much more like the new analog meter than the old analog, which means I've been using a bad calibration file for a long time.

Sonnie, do we have permission to post the file contents here?

Absolutely! We are not copyrighting calibration files... Distribute as you wish.


Is it a fact that Radio Shack has only had one model number for the digital meter? I need to make note of this if so... I couldn't seem to verify it one way or another.

I can only say that it's been the same since 2004, and that the digital looks just like the old analog (different insides, though).

I see that the calibration files only go up to a bit over 100 Hz. Did you compare response above that at all? If I recall, the new meters are reported to have a difference HF response over about 8 kHz.

We didn't compare above 100hz since we are mainly using this for sub measurements and we were allowing C-weighting to pick up the slack from 100-200hz. I personally use my calibrated ecm for full range for the few times I've looked at it. Plus it's better to use a flat response when calibrating and it would be impossible for me to get anywhere remotely close to flat full range in my room. It's ugly above 100hz with all the zigzagging.

It's really only the response below 100Hz where any of the RS meters diverge from the C weight standard with any significance. Above that we allow REW (Room EQ Wizard) built in C weight compensation to take over. If a meters calibration file is loaded, the C weight compensation is in effect outside the frequency range limits of the calibration file. In the case of sonnies calibration files, the C weight compensation merges and takes over at >~105Hz. Certainly the area of concern is from 10Hz to 200Hz for equalizing a subwoofer. Any readings with an RS meter above 10KHz are questionable.

Be cautioned that using measured data (as we have done) is a dubious method to derive a calibration file for a meter. We simply decided that it would be a heck of a lot more accurate than what's available - which until sonnie began testing was a single crude low resolution file for all the three types of RS meters.

Using the calibration files that sonnie has posted on his forum for the three meter types, measuring against a professionally calibrated ECM microphone, provide results that are quite close and acceptable for sub EQ. Using the best sweep to sweep results there are always anomolies which can be attributed to anything from element positional differences to someone moving around the room or scratching their nose. Either way, the calibration results are quite close.

brucek

These are for the current analog meter used as a microphone.

10 -7.29
11 -6.85
12 -6.35
13 -5.70
14 -5.10
15 -4.60
16 -4.30
17 -4.00
18 -3.70
19 -3.55
20 -3.25
21 -3.15
22 -3.00
23 -2.90
24 -2.80
25 -2.65
26 -2.58
27 -2.45
28 -2.25
29 -2.09
30 -2.01
31 -1.91
32 -1.80
33 -1.71
34 -1.55
35 -1.40
36 -1.27
37 -1.15
38 -1.07
39 -1.00
40 -0.98
41 -0.93
42 -0.86
43 -0.75
44 -0.69
45 -0.63
46 -0.59
47 -0.55
48 -0.50
49 -0.49
50 -0.46
51 -0.46
52 -0.45
53 -0.46
54 -0.47
55 -0.47
56 -0.48
57 -0.49
58 -0.50
59 -0.51
60 -0.53
61 -0.55
62 -0.57
63 -0.59
64 -0.61
65 -0.63
66 -0.65
67 -0.67
68 -0.68
69 -0.72
70 -0.75
71 -0.77
72 -0.79
73 -0.79
74 -0.80
75 -0.82
76 -0.83
77 -0.84
78 -0.86
79 -0.85
80 -0.86
81 -0.87
82 -0.88
83 -0.91
84 -0.94
85 -0.94
86 -0.93
87 -0.92
88 -0.92
89 -0.91
90 -0.90
91 -0.89
92 -0.88
93 -0.87
94 -0.86
95 -0.85
96 -0.84
97 -0.83
98 -0.83
99 -0.82
100 -0.82
101 -0.81
102 -0.80
103 -0.78
104 -0.76
105 -0.73

These are for the current digital meter used as a microphone.

10 -5.66
11 -5.25
12 -4.66
13 -4.20
14 -3.80
15 -3.50
16 -3.30
17 -3.10
18 -2.90
19 -2.70
20 -2.50
21 -2.30
22 -2.10
23 -1.95
24 -1.80
25 -1.60
26 -1.40
27 -1.15
28 -1.00
29 -0.90
30 -0.75
31 -0.65
32 -0.55
33 -0.44
34 -0.33
35 -0.22
36 -0.11
37 0.00
38 0.08
39 0.15
40 0.22
41 0.34
42 0.36
43 0.41
44 0.45
45 0.49
46 0.52
47 0.55
48 0.59
49 0.60
50 0.62
51 0.64
52 0.64
53 0.64
54 0.63
55 0.62
56 0.60
57 0.59
58 0.56
59 0.53
60 0.50
61 0.48
62 0.44
63 0.41
64 0.39
65 0.37
66 0.36
67 0.34
68 0.30
69 0.28
70 0.27
71 0.30
72 0.32
73 0.34
74 0.34
75 0.35
76 0.36
77 0.37
78 0.38
79 0.39
80 0.42
81 0.42
82 0.43
83 0.43
84 0.43
85 0.43
86 0.43
87 0.44
88 0.45
89 0.45
90 0.46
91 0.46
92 0.47
93 0.48
94 0.48
95 0.49
96 0.49
97 0.50
98 0.50
99 0.50
100 0.50
101 0.50
102 0.50
103 0.50
104 0.50
105 0.50

These are for the now-discontinued analog meter used as a microphone.

10 -27.00
11 -24.15
12 -21.91
13 -20.13
14 -18.61
15 -17.33
16 -16.15
17 -14.92
18 -13.85
19 -13.01
20 -12.38
21 -11.66
22 -10.79
23 -10.00
24 -9.33
25 -8.80
26 -8.41
27 -7.99
28 -7.50
29 -6.99
30 -6.50
31 -6.03
32 -5.60
33 -5.23
34 -4.89
35 -4.58
36 -4.29
37 -4.04
38 -3.83
39 -3.64
40 -3.46
41 -3.31
42 -3.16
43 -3.01
44 -2.88
45 -2.74
46 -2.60
47 -2.46
48 -2.35
49 -2.26
50 -2.21
51 -2.18
52 -2.15
53 -2.11
54 -2.04
55 -1.98
56 -1.95
57 -1.92
58 -1.88
59 -1.84
60 -1.80
61 -1.77
62 -1.73
63 -1.71
64 -1.71
65 -1.72
66 -1.70
67 -1.67
68 -1.64
69 -1.60
70 -1.56
71 -1.52
72 -1.47
73 -1.41
74 -1.34
75 -1.26
76 -1.19
77 -1.12
78 -1.07
79 -1.02
80 -0.98
81 -0.94
82 -0.89
83 -0.85
84 -0.80
85 -0.75
86 -0.71
87 -0.66
88 -0.62
89 -0.59
90 -0.55
91 -0.52
92 -0.50
93 -0.47
94 -0.43
95 -0.40
96 -0.36
97 -0.34
98 -0.32
99 -0.30
100 -0.27
101 -0.25
102 -0.23
103 -0.20
104 -0.17
105 -0.14
106 -0.11
107 -0.07
108 -0.03

An enterprising person might merge these corrections with those posted by Ikka earlier in the thread to get a better full-range set. Unlike the old widely available calibration file, his corrections are somewhat lumpier and appear to map to the response curve measured by Sonny and friends.

Isn't it amazing how the old original "correction" values which have been circulated far and wide for years were so totally OFF!!! Many thanks to Sonny, Bruce, and friends for doing us all a huge service with much diligent research and testing to finally publish something closer to the TRUTH and especially for being tailored to the respective types/models of RS meters!!

Thats for all the work.

Its refreshing to see that the new meters are getting more accurate. I have been using the digital for a while now and am actually very pleased with the unit.

-Eli

Was only one of each type measured? Wondering about the unit-to-unit variation.

Thanks

lol, looks like all my hand drawn "calibrated" curves need to be redone.

I only had one of the older model analogs... but I had two of the newer model analogs and two of the digital meters. They were very close to each other in response as you can see from the graphs in the test thread. Even so, those two responses were averaged (the two for the analog and the two for the digital) and then the corrections were set to the average. That should keep us fairly close all around... I'd day you will be within +/- 1db.

If we consider the ATI meter (supposedly took over the meter manufacturing - although I'm not 100% on this - I simply think they did) and their claims of +/- 2db from 32hz on up... this coincides with what we found at least from 30hz up to about 120hz. So it does appear they are getting more standardized/uniform in their manufacturing process and maybe using better components/caps, etc. This would indeed seem to indicate that there's very little variance from meter to meter. Unfortunately I can only confirm this between two meters of each, analog and digital. Even the analog and digital were very close to each other... the minor variances were below 30hz.

I must have missed this part. How do I know if I have an older analog model or a newer one?

Look at the one currently offered on the Radio Shack web site. If it looks like that one, it's "newer". If it looks more like a brick, it's "older".

Or if you were really lazy, you could use the catalog numbers posted above to match to the number on the back of your meter.

Thanks for the reply - I have the newer model.

So - I measure the response at 20hz with a test tone then I have to subtract 3.25 to get the true response? Or do I have to add 3.25. (Using your table above)

Thanks

Your measurement is down 3.25db at 20hz so you will add this to your measurement.

Add negative numbers and subtract positive numbers.

Excellent work guys!

I hate to confuse the situation but I have the older analog model 33-2050 purchased in Canada many years ago and have compared it to my Behringer ECM800 (not calibrated) and assuming the ECM is reasonably accurate out of the box my older RS meter measures fairly accurate and the new table won't apply.

My setup is like this: I'm running the ECM into the metering of a DEQ2496, using the line-out of the RS meter into ROOM EQ Wizard, and of using the RS metering needle as well. Both meters set to C-weighting. 1/3 octave tones provided by Stereophile Test CD2. I planned to use 1Hz increments from ROOM EQ Wizard but ran out of time.

I ran the measurements a few times the first time through with a battery that's been in the unit for awhile and these first measurements had me about 5dB out at 20 Hz from where they increasingly improved improved.

On the next two runs I installed a new fresh battery and the results improved, the corrections I calculate are referenced to the ECM8000:

Line out
200 -.8
160 -1.2
125 -.5
100 -.4
80 -13.6
63 +.7
50 +3.7
40 +.7
31.5 +1.0
25 +1.0
20 +3.5

Needle
200 -1.1
160 -1.2
125 -.5
100 -.8
80 -1.8
63 -.9
50 -1.4
40 -1.0
31.5 +1.2
25 +.7
20 +2.1

I'm a little surprised how close the two meters are, the meters needle being more accurate than the lineout which shows a system problem at 80hz.

So the obvious question: is it possible an uncalibrated ECM8000 is out 10 dB at 20Hz? Anyone have there's calibrated and wish to comment?

Also for Sonnie, how many older model 33-2050 analog meters were sampled to obtain your correction values and was it/they outfitted with a fresh battery?

On a side note, when compared to the ECM8000 my RS meter is most accurate reading 60 dB and steadily becomes less accurate as dB levels increase being out about 2 dB at 100 dB.

I think some of your difference may be C-Weighting. The meter's RCA output is not weighted. You'll need to factor that into your calculations above.

My ECM is -1.8db at 20 hz... -5.33db at 10hz. This is pretty close to what I've seen from other calibrated ECM's. Some may have been a little different at 20hz but not 10db.

I used a new battery because when I started my battery was dead and had to get my wife to pick me up a new one. I only tested the one older meter but if you look back at the other test done by others here, their corrections were pretty close to mine except at 10hz and 12hz, and I think the other one had an error there. Here's what I posted on that earlier:


It's also important that when you compare mics that you are using sweep or tones as flat as you can get. On REW you will need to set the Pre-ref window to 0.25 Tukey prior to taking measurements and the Pre-ref window to 125ms after measuring. All this will affect your results. Plus we ran a sweep and smoothed the response to remove the spiky anamalies.

Hi Sonnie,

I suppose first it would be important to understand the purpose of the "calibration files". I took it to undertand that files were to compensate for inaccuracies of the RS meter when compared to an equally weighted reference meter and this is what I performed, apples to apples.

If, however, the "calibration" files are actually c-weighting compensation/error correction files intended to adjust for inaccuracy and c-weighting from the RS values then this is slightly different. If I have time tonight I'll rerun the test to see if the weighting compensation factors are similar, I suspect they'll be close. Perhaps I missed some new info but I thought it was determined the old RS analog meter has c-weighting applied to the line out, the digital meter does not.

As a general question, it seems faulty to not use c-weighting when calibrating speakers, the intent is to end up with a flat response as heard by your ears not a piece of electronics. Unless the test tones are weighted I don't see any value in not using c-weighting. What the hell am I missing here?

It really doesn't matter whether C-Weighting is applied or not... the end results need correction values to make the mic in the meter read flat.

I cannot say for sure this is your difference between the needle and the RCA output, but it seems logical... I'm not why else there would be a difference. Not why else they would be different. But irregardless... when we plug into the RCA output of those RS meters, we do not get a flat response and (as you stated) the CV's compensate for this.

This whole thread is frightening.

I ran one of the two "Type 0" acoustic laboratories in the country for three years, so I know a thing or two about acoustical measurements.

I don't have a lot of time to get into all the details, but the short of it is Radio Shack and other low cost SPL meters should not be used for Acoustical measurements (response or otherwise) under ANY circumstance. They are specifically designed for UNCALIBRATED, BALL PARK "how loud is this" type of measurements (about 50Hz-5Khz). Any real piece of measuring equipment is labeled either Type 1, or Type 2, or if you get specialized equipment, even Type 0. Even IF these inexpensive meters declare they are Type 2 for "A" weighted measurements, they can still be very inaccurate at low and high frequencies. These meters vary so much from one to the next that a calibration file for one is meaningless for another. In addition, they are not good for measuring above about 105dB. This has to do with poor electronics, but more importantly, the super poor quality of the mics they use.

I know most of you can not afford the Type 0 equipment we use here at Acoupower. But, there are some reasonable alternatives:

1) The AudioXpress site has a good microphone for a reasonable price.
2) Many companies offer PC Card based test and measurement systems that are reasonably accurate and only a few hundred dollars.

The bottom line is there are several ways to get a pretty good test and measurement setup for about $500.00.

Again, please visit the AudioXpress site for more information.

Hello Carlos,

I here where you are coming from, but most people who spend 100 bucks on a BFD aren't too interested in spending 4-500 bucks on a mic setup. While the RS meter isn't perfect, it will be within a few db of accurate. From meter to meter they run about 1-2db in variance, so that's not bad. Most folks won't notice a difference in a couple of db. For the most part we use these in the 20 - 200hz range at about 75-85db. For our purpose they are fine. If someone wants to spend the money to get within .5db of accuracy then there are plenty of mics available for less than 4-500 that will do that too. Remember that most of us are simply the average homeowner and most of us don't want to spend as much or half as much for a mic setup as we did our sub.

Sonnie,

Personally, I wish what you were saying were true.

I have experience measuring these units with laboratory grade equipment. Unfortunately, at the very lowest and highest audio frequencies, these meters can vary +-10dB. Actually, near 20kHz, they can vary even more. They are only REASONABLY accurate from about 50 to 5kHz.

That means a very "hot" meter at 20Hz might read 20dB higher than a very low meter. That is much more than a dB or two.

Again, the bottom line is using these meters for ANY kind of SPL vs. Frequency measurements is outside of their intended application.

This is why the first line in my first post on this thread was "I find this whole thread frightening".


Carlos

ATI claims accuracy from 32hz to something another. And even if it's only 32hz to 5khz... that's pretty good.

There's been several people test these things in the real world environment and we've just never seen 10db difference at 20hz from the same model meter to meter... even calibrated against different fairly good quality mics. If our response is off 10db at 20hz, we'll probably be able to notice it... and say... this don't sound right. Plus, for the most part we are trying to tame the bad peaks/spikes and most of those don't occur at 20hz... most are between 30-60hz... so I believe we can get relatively close.

Hey... most of us are simply doing the best we can for the least $$$ we can.